Complex Network Model of Loss of Privacy in Social Network Contagion

Abstract

Social networks whether offline or online influence our lives significantly. Loss of personal information is one of the key issues in securing any social network. Online social networks are more exposed because they hold information about users on servers. They are considered to be vulnerable for privacy exploitation. To the best of our knowledge, there is no attack which evaluates change of connections. While there are other online social networks attacks, change of connections attack is proposed. The connections that exists in social networks are exploitable. A hacker may change friends connections in online social networks without the knowledge of system administrator. Also, currently there is no mechanism which allows for a scalable solution to detect information tampering in online social networks. To understand the dynamics of information diffusion in social networks studies were conducted. First, an empirical study was conducted over 3 online social networks data sets. These datasets were collected from online social network, Twitter. These datasets were collected to analyze the importance of centralities. To further validate the idea, 4 already published offline social networks data sets were next taken besides including a random network for comparison. Furthermore, a cryptographic mechanism was proposed that combines centralities and applies a cryptographic hash algorithm to detect any changes in networks. Based on online case studies, it was discovered that centralities play an important role in networks. These case studies demonstrated that centrality measures portray importance of nodes. Also, centralities are useful in measuring information diffusion in networks. After further investigation, the empirical analysis of offline data sets showed that different centralities have different impact over networks. Thus, an individual centrality might always not be a true judge . This resulted in combining multiple centralities to be used together in the proposed solution. The proposed tamper-evident mechanism was evaluated on a comprehensive social network case study. The successful application of the mechanism was demonstrated by the detecion of even the most minor changes in a network allowing the system administrator to become aware of such unauthorized access irrepective of the complexity or number of nodes in the social network.